In progress This essay discusses the recommended way for admins to close AFDs. The views contained in this essay will not reflect every admins' approach and preferences but it is hoped that through editing a wider consensus to how to do this should emerge.

Closing AFDs can be a difficult and stressful area for administrators to work in, not just because of the difficulty of weighing competing arguments but also because AFD is caught right in the middle of an internecine battle between those that want to keep crappy articles come what may and those vandals who hate content and want to delete everything. Okay the descriptions are a joke but demonstrate how polarised the divide is becoming. Of course there is a middle way of just closing everything as "no-consensus" or taking the easy way out of counting noses in the discussion but this generally doesn't adhere to what we consider to be a consensus in Wikipedia.

Consensus in the Wikipedia model is the process of weighing arguments against policy to decide which arguments are the most valid and therefore win the discussion. Consensus is not reached by counting the number of votes but it would be foolish to decide that one well argued delete vote was more worthy then 10 equally well argued keep votes because you personally feel the delete vote was slightly more valid. The art of closing discussions is to be able to weigh the competing arguments and to recognise what arguments are based on policy and what arguments are not. Ah, but its not as easy as that, because a local consensus does not over-ride a global consensus on how to deal with particular types of article. So a discussion that clearly shows a numerical consensus to delete an article on a primary school is actually the wrong outcome as there is a global consensus to redirect/merge these articles into the pages of the school district/local education authority. Also a local consensus to keep something because its really cool and interesting isn't going to fly if there are no reliable sources because the global consensus is that we expect articles to be verifiable and notability requires sources.

There has also been a recent trend for discussions not to centre around policy and this makes closing very difficult as whole sections of the discussion might be worthless in helping you understand the policy based consensus. Discussions like this are good candidates to be relisted but you would be advised to give the participants some direction on what the closing admin will need to find the consensus next time round. If this doesn't work then the discussion is almost certainly doomed to be closed as "no-consensus"

Now for the slightly controversial discussion. It's also important to understand about voting blocks and how they might affect the outcome of AFDs. Listing an article for rescue for example is likely to lead to a number of votes by members of the article rescue squadron who generally vote to keep content as its the kind of project that tends to attract hard-core inclusionists. Also nationalistic articles often attract participants from those areas and discussions often fall into a nasty argumentative battleground where the consensus is lost in recriminations and counter-claims. The important part here is that most AFDs get closed around notability arguments and having an influx of users knowledgeable about the subject or adapt in finding sources is useful. Just ignore any arguments not based in policy, ban/block/mute anyone attacking other editors and just concentrate on the sources provided and whether they cut the mustard for notability. If there are decent sources then the consensus is clear and, if there are not, then the consensus is also clear.

After a while working on AFDs its possible to get a clear feel for the way that regular AFD participants will go and when they vote different to the norm it is worth reading their comments extra carefully as these are the ones that often shed the most light on the overall consensus.

This directly from the notability guideline

:If a topic has received significant coverage in reliable sources that are independent of the subject, it is presumed to satisfy the inclusion criteria for a stand-alone article or stand-alone list.

• "Significant coverage" means that sources address the subject directly in detail, so no original research is needed to extract the content. Significant coverage is more than a trivial mention but it need not be the main topic of the source material.^[1]
• "Reliable" means sources need editorial integrity to allow verifiable evaluation of notability, per the reliable source guideline. Sources may encompass published works in all forms and media, and in any language. Availability of secondary sources covering the subject is a good test for notability.
• "Sources,"^[2] for notability purposes, should be secondary sources, as those provide the most objective evidence of notability. The number and nature of reliable sources needed varies depending on the depth of coverage and quality of the sources. Multiple sources are generally expected.^[3] Multiple publications from the same author or organization are usually regarded as a single source for the purposes of establishing notability.
• "Independent of the subject" excludes works produced by those affiliated with the subject including (but not limited to): self-publicity, advertising, self-published material by the subject, autobiographies, press releases, etc.^[4]
• "Presumed" means that significant coverage in reliable sources establishes a presumption, not a guarantee, that a subject is suitable for inclusion. Editors may reach a consensus that although a topic meets this criterion, it is not appropriate for a stand-alone article. For example, such an article may violate what Wikipedia is not.^[5]

A topic for which this criterion is deemed to have been met by consensus, is usually worthy of notice, and satisfies one of the criteria for a stand-alone article in the encyclopedia. Verifiable facts and content not supported by multiple independent sources may be appropriate for inclusion within another article.

Sources are generally the key issue for finding consensus. GNG says that non-notable articles will be deleted. Notability mostly derives from the sources and even when notability is assumed by the application of a sub-guidelines those guidelines have been reached by working out whether meeting them means that sources are likely to exist. This means that a very thorough search for sources that includes trips to a library and/or scouring on-line sources is a basic requirement for deleting an article. Actual sources that pass muster always trump a majority argument that they don't exist. Assertions of sources will always fail if the delete side show that they haven't found any by a thorough search and no actual sources are adduced in the discussion.

One problem for a closing admin is assessing the sources. Should they make their mind up on them (bearing in mind that most admins are far more experienced in policy then the average user) or should they go with the discussion. What if the discussion doesn't discuss the nature of the sources but simply descends into "Oh no they aren't/Oh yes they are"? Admins need to avoid supervotes but they are allowed to find on matters of fact. If the source provided is clearly inadequate then you can rely on your own interpretation but if the source is questionable then a discussion of its merits is a much safer way to find the consensus. Relist and direct and, if necessary, ask some RS experts to contribute to the discussion to help with resolving the issue. If, having tried this, the source is still unclear and the AFD has not come to a clear outcome on the source then you need to close as "no-consensus".